The problems are quite varied. There are difficulties with certain kinds of multimedia technologies. The most obvious example is Flash. It's still largely inaccessible.
Most screen readers still aren't capable of supporting it and most developers who use Flash, particularly on web sites, don't understand how to create it in an accessible fashion.
In other respectsm a lot of multimedia is obviously visual in content so if it doesn't have an audio description to describe the content that's happening, then vast amounts of action on the screen can pass by completely un-noticed.
It's difficult to make Flash accessible because Macromedia are still working on permitting the developer to do that.
They've done some incredible work in the past 2 or 3 releases of a Flash development environment, but at the end of the day Flash is an almost expressly visual medium. It's an animated medium.
So there are problems from a development perspective. There are also problems because it's a visual medium in just understanding what the rest of the world is looking at. Buttons may be acknowledged as being a button, but what the text that everybody else sees on that button is often utterly impossible for me to tell.
If there's any just straightforward, you know, animation of images and bits and pieces, there's no way of translating that into any recognisable audio feedback.
Screen readers can't translate graphics into audio information so there's very much a big gap in the way between my technology and my screen reader technology and the possibilities of the Flash technology.